In order to demonstrate challenges to conventional wisdom (Aldrich 1980a, b; Bartels 1985 1988; Orren and Polsby 1987), this article develops several forecasting models of the presidential primary vote to compare to a baseline model of the aggregate primary vote (APV) that uses pre-primary and New Hampshire primary data. The models indicate that candidates' Gallup poll position and cash reserves are significant positive predictors of a candidates' primary vote share, though there are differences between forecasting models of the primary vote in Democratic and Republican nomination campaigns. Parallel models incorporating results of the New Hampshire primary improve the predictive power of the baseline model, indicating that the bellwether primary has a "correcting" effect on the relative standings of some candidates seeking a presidential nomination. This effect is substantially greater for Democrats than for Republicans.
The reform movement of the early 1970s shifted power in presidential nomination selection away from party elites during the nominating conventions to primary and caucus voters (Ranney 1975; Kirkpatrick 1978; Ceaser 1979). In Aldrich's (1980a) framework, the reforms created a presidential nomination process where candidates, party activists, and primary voters make the key decisions "before the convention." The conventional wisdom that emerged held that early momentum generated by performing well in bellwether states like Iowa and New Hampshire was critical to winning the nomination (Bartels 1985, 1988; Orren and Polsby 1987). Where Aldrich argued that critical period of the nomination campaign occurred between January of the election year and the nominating conventions held in mid-Summer, this research contends that the critical phase of the nomination campaign is now "before the primaries."
The initial openness of the post-reform system narrowed significantly since the 1970s. Today, states "frontload" the calendar by holding their primaries on earlier dates. By March 15th of 1976, only 15 percent of the primaries (four states) were held, compared with 60 percent of the primaries (26 states) by March 15 of 2000.1 In addition, the cost of campaigning for either party's nomination is growing exponentially. In 1976, no candidate raised more than $3.5 million prior to the primaries and in 2000 no candidate remained in the race until the first primaries with less than $3.5 million. The more compressed primary schedule and increasing campaign costs make the competition for resources during the pre-primary period more important to winning presidential nominations. Campaigns can no longer hope for momentum from early primary performance to generate the resources needed to compete in the middle and later primaries; they simply lack the time to raise enough money or to build national organizations once the voting starts. Nor can candidates rely on free media coverage to compensate for what they lack in resources-in fact, network news coverage of nomination politics is declining (Steger 2002) and the news media are increasingly critical of presidential candidates (Owen 1991, 1997; Steger 2001). These developments accelerate and increase the importance of pre-primary campaign activity.
The movement toward earlier fundraising, organizing, and active campaigning appears to reduce the effect of early primaries and caucuses on the nomination contest (Adkins and Dowdle 2001b). Since 1980 the candidates identified as frontrunners in January won the nomination despite losing in either or both the Iowa caucuses and the New Hampshire primary. Rather than giving dark-horse candidates the opportunity to "break through," the early contests represent "bumps in the road" to the party nominations (Adkins and Dowdle 2004).
This does not necessarily imply the emergence of a new era of presidential nomination politics. Each time a new system materializes, it opens up or democratizes the process in lhat more people can participate in the choice of who is nominated, and more people can run for the nomination (e. …